The present invention relates to an information processing device, comprising a computer device, and a storage device for storing data used by the computer device, wherein data is relocated between a plurality of storage devices in accordance with a change in the characteristics of the data with the passage of time.
There are disk array systems which achieve high-speed operation and high reliability by providing a plurality of disk devices in an array fashion, in a single two-dimensional storage system for an information processing system. The disk devices have distributed access and are operated in a parallel fashion, thus providing redundancy for the stored data.
A disk array system provided with redundancy is assigned various RAID levels, from RAID level 1 to level 5, according to the redundancy structure (for example, see the below-listed Non-Patent Reference 1), and data transmitted from a host computer connected to the disk array system is located in a plurality of disk devices, in accordance with the RAID level.
In a disk array system, the logical storage region that is recognized when the system is accessed by a host computer is associated with a physical storage region indicating an actual storage region in the disk device, in such a manner that the host computer is not aware of the fact that the data is stored in a distributed fashion in a plurality of disk devices.
When constructing a disk array system of this kind, in order to achieve optimal cost performance, disk devices of differing storage capacities and differing characteristics may be used in combination. In a case of this kind, desirably, the data transmitted by the host computer and located hypothetically in a logical storage region is stored in the physical storage region provided by the most suitable disk device, on the basis of indicators relating to access frequency and access patterns, and the like.
In a system which combines disk devices of different characteristics, the method used for allocating the data to the physical storage region provided by the most suitable disk device involves, for example, technology whereby the data access frequency is monitored and data is moved to the most suitable physical region on the basis of the access frequency (see, for example, the below-listed Patent Reference 1.)
In the technology disclosed in Patent Reference 1, a threshold value is set previously, the frequency of access to the data is monitored, and if the access frequency has exceeded the threshold value, then the data is moved to a physical storage region provided by the disk device operating at a higher speed. Furthermore, it is also possible to relocate the actual logical storage region itself, wherein the disk array system monitors the conditions of the access load to the respective logical storage regions from the host computer, and determines the details of the relocation in such a manner that the data is optimally located after relocation, in accordance with the results of the load monitoring. When the logical storage region has been relocated, the association between the logical storage region and the physical storage region is changed to the physical storage region after relocation.
Furthermore, there is also technology for performing relocation wherein the use status of a disk device corresponding to read/write operations from the host computer is gathered, the disk access is predicted from the information thus gathered, and the most suitable disk device for locating the data is determined (see, for example, the below-listed Patent Reference 2).
Moreover, in a document relating to SMI-S (Storage Management Initiative Specification) under deliberation by the SNIA (Storage Networking Industry Association), which is a storage industry group, technology for a storage system consisting of a disk array system is described, wherein a logical storage region is assigned to a physical storage region provided by a disk device, in accordance with previously determined “hints” which indicate the access characteristics for the data (see, for example, the below-listed Non-Patent Reference 2). The storage system conforming to SMI-S is equipped with functions for selecting a physical storage region on the basis of a “hint” and assigning a logical storage region to the same.
(Non-Reference Patent 1) D. Patterson, G. Gibson and R. H. Kartz, “A Case for Redundant Arrays of Inexpensive Disks (RAID)” (ACM SIGMOD, June 1988, pp. 109-116)
Reference Patent 1: Japanese Patent Laid-open No. (Hei) 9-274544
Reference Patent 2: Japanese Patent Laid-open No. 2002-82775
(Non-Patent Reference 2) “SMI-S Specification Public Review Draft” p. 157 (SNIA, 2003)
(Problems to be Solved by the Invention)
A storage system which moves data in accordance with various circumstances in the above-described manner provides a method for achieving optimal location of the data.
However, in the technology disclosed in Patent Reference 1, as described previously, the logical storage regions are monitored, and relocation of data starts at the time that the load has exceeded a threshold value. Since relocation of the data takes time to carry out, in cases where there is a sudden increase in disk access, the data relocation is not completed in time, and, hence, the benefits of relocation cannot be obtained. Moreover, since the data relocation process generates a large number of disk access operations, then while relocation is being executed, the disk access operations required for data relocation are added to the normal disk access operations, and, hence, the overall efficiency declines.
In the technology disclosed in Patent Reference 2, the future data access characteristics are predicted on the basis of past disk access characteristics, and, therefore, relocation is carried out in advance, in accordance with the predictions made. However, it is only possible to respond by means of predictions based on past history in cases where the same change characteristics are generated in a continuous fashion, at uniform time intervals. Even if the change characteristics are the same, for example, in many cases, disk access occurs in an irregular manner over time, and, therefore, it is not necessarily possible to predict the most suitable time for relocation of data, on the basis of past disk access characteristics. Therefore, in the technology disclosed in Patent Reference 2, data is not necessarily relocated in an optimal fashion, as desired.
On the other hand, there is also the method whereby “hints” defined according to SMI-S are provided when creating a new logical volume. However, the SMI-S indicates how existing logical storage regions are to be used in the future, and it does not consider processing for determining the time at which a logical storage region is to be relocated, or the destination to which it is to be relocated.